GRAPHIC RECORDING OF SPEECH VIBRATIONS 
95 
from the prism Pri to the open space 00. If, at the same time 
voice or other sound waves are passing through the free space 00 , 
the refraction of R will vary slightly according to the successive 
rarefactions and condensations caused by the varying phases of 
the sound waves passing through 00 , thus shifting the ray or 
light-lever into different slightly divergent paths RR and RR\ 
This divergence is increased by the lenses or mirrors, thus pro- 
jecting on PP an enlarged sinusoidal record of the sound waves 
passing through 00 . 
Two modifications of the second prism Pr 2 will probably be 
found necessary in actual practice. First, it would, if unmodi- 
fied in form, tend to neutralize the divergence of the light-lever 
R^ effected by the varying air densities in the opening 00 , since 
the refractive indices of the two prisms Pvi and Pr^ would be 
reciprocal to each other for every density of air in the opening 
00, while the air density in both sound proof chambers SC and 
S'C' remains constant. To avoid] this neutralizing effect, it 
will undoubtedly be necessary to concave either the first surface 
or both surfaces of the second prism Pr^ at the point or small 
area where the light-lever R traverses it, the first center of curv- 
ature or concavity lying either at the point where the light- 
lever R emerges from the first prism Pri or within the open 
space 00. This would in effect introduce a minute concave or 
bi-concave lens in the second prism Pn at this point. Again, the 
second prism Pr^ may have to be doubled, with a dead air space 
between its two parts, in order to preserve the sound proofing of 
the second chamber S'C'. 
In form A (fig. 4), the arrangement of lenses and focussing 
is the same in principle as that of form A of the sonoscope, 
shown in figure I. This plan would probably necessitate the 
distance RL 2 (from the first prism to the fiber lens) to be 
relatively very great in order to allow the very slight divergence 
of the paths RR and RR' to become sufficient for the lens L 2 L 3 
to deal with it. How great this difficulty will prove to be in 
practice can be determined by trial. 
In form B (fig. 5), two convex glass-tube mirrors Mi and M 2 
replace the lenses L 2 L 3 and L 4 of form A. The focussing follows 
